biography wireless data logger networks · 12/11/2011 · wireless data logger networks by javier...
TRANSCRIPT
Wireless Data Logger Networks
by
Javier González
My name is Javier González, I am 24 and I started my 3-year Ph.D in N o v e m b e r 2 0 1 1 a t t h e I T University of Copenhagen. My Ph.D is partly funded by the FP7 infrastructure project INTERACT, under the supervision of associate profes sor Ph i l ippe Bonnet . xxxxxxaa
CONTEXT
Data Management & Time SeriesHow do we integrate a sensor net with a DBMS?
• Flexible Data Loading How to deal with data sets coming from diverse sources and structure?
DBMSdata
data data
• Enforce Usage ControlDefine some kind of usage control policies for data generated in the arctic ecologists community is an open issue. Designing an infrastructure for enforcing these policies is also an open issue.
IT UNIVERSITY OF COPENHAGEN Javier González - [email protected] supervisor: Philippe Bonnet April, 2012
Ph.D Forum - IPSN'12
After obtaining both my Bachelor and Master's at the University of Valladolid I moved to Denmark, where I carried out my Master's Thesis and stayed as a Ph.D.
Data Forwarding at High-BaudratesSustaining high speed serial data forwarding is a problem for EPIC and TELOS motes running TinyOS. For a mote to support a 115200 baud data rate, the UART must be available every 0.07ms. Since (i) the UART is interrupt-driven and since (ii) there are components in TinyOS that either are non-interruptible or disable interrupts, this cause the UART to miss interrupts and thereby data.
Stand-Alone data loggers
NetworkedSystem
Integrating stand-alone digital devices not affected by the MEMS revolution, and equipped with a high-speed serial interface (115200bps) into a networked system has proven to be a challenge for commercially available EPIC and TelosB based systems. We refer to this problem as Wireless Serial Data Transmission (WSDT).
RS-232
Monolithic, Bulky, E x p e n s i v e a n d Energy Inefficient sensors
115200bps
Biography
We focus on how to handle vast amounts of diverse, small data sets. Also, we aim to provide a tool to organize the management, publication and archival of these data sets. This problem was originally pointed out by Jim Gray. Access and usage control are necessary to enforce existing data policies.
DataSet 2
DataSet 3
DataSet 6
DataSet n
DataSet 7
DataSet 4
DataSet 5
DataSet 1
DataSet 6
DataSet 4
DataSet 5
DataSet n
Efficient Data Loading, Access Control & Usage Control
COLLABORATIVE DATA INFRASTRUCTURE
Publica-tion
DataSet 7
DataSet 2
DataSet 3
DataSet 6
DataSet 4
DataSet 5
DataSet 1
DataSet n
OTHERDATA+ +
Ownership
Author(Researcher)
Research+
Feedback
UsageControl
Recognition
Researcher
Researcher
Field Work
Collected Data
DataSharing
DataSet 2
DataSet 1
DataSet 7
DataSet 3
Data Sharing (INTERACT)
Who accesses data? - Access Control
H o w d a t a i s used? - Usage Control
PROBLEMS ONGOING WORK
INTERACTINTERACT i s an EU-founded, FP7 infrastructure project. Its main goal is to build capacity for research and monitoring along a network of 33 terrestrial field bases in northern Europe, Russia, US, Canada, Greenland, Iceland, the Faroe Islands and Scotland.
www.eu-interact.org/
Equip data-loggers with wireless capabilities (INTERACT - WP5)
Reliable, precise, stand-alone sensor + data loggers
RS-232
Serial Data
RS-232
Send CommandCommand
Send Packet 1
ACK
...Send Packet n
ACK
Poke - Wake Up
Answer
End Comm
for (i=0;i<n;i++) send_Packet(i);
t
Act as a RoutingDevice
115200bps
In the context of INTERACT (Work Package 5 - Virtual Instrumentation), our initial goal is to equip data loggers with wireless capabilities to forward serial data at high baud-rates (115200bps) without data-loss (sensible data) in single-hop and multi-hop topologies. Adding routing capabilities while keeping up this high rates pushes current platforms to their limits.
SerialData
Base Station
How should a wireless data logger network be programmed? Who is programming it? Which are the needed abstractions? How generic should a data logger network Programming Framework actually be?
There is a mismatch between the current incarnation of TinyOS (genericity & hardware independence) and actual needs of sensor net programming. Including a lack of an agile data forwarding capability among serial interfaces, Radio and FLASH due to shared SPIs. Arduino motes + XBee modules can easily achieve this.
Hardware Design & Programming Frameworks
Arduino MegaTelosB & EPIC
µC
Serial
Serial
Serial
Serial
UART 0UART 1UART 2UART 3
XBee
RS-232
FLASH
SPIµC
FLASH
Radio
Serial
SPI
RS-232 *
We revisit the importance of hardware design on embedded network programming.
TelosB & EPICArquitecture
*Different FLASH chip for Telosb (m25p80) and EPIC (AT45DB041D )
RS-232
µCFLASH
Radio
Serial
UART 0
SharedSPI
UART 1
M430f1611
P44
P42
*
Scientists are more interested in precision than in spatial coverage. Reliability and Robustness are top requirements.
Internal Cyclic Buffer (RAM)
...
UART
Clock
FLASH
Radio
µCRS-232115200bps
115200bps
User
Provider
Interface
Commands
Events
Image taken from TinyOS Programming
Also, the event-based nature of TinyOS combined with atomic statements obscures the real processes being taken inside the internal components of the motes.
WSDT in TelosB+TinyOS leads to unacceptable data-loss rates and mote hanging.
We are pushed to develop smart software strategies in order to make up for hardware deficiencies
Impact the Design of monitoring components
Generalizing from the WSDT example, we investigate the interplay between hardware design and programming framework for specific classes of network embedded systems.
µC
Serial
Serial
Serial
Serial
UART 0UART 1UART 2UART 3
XBee
RS-232
FLASH
SPIµC
FLASH
Radio
Serial
SPI
RS-232 * +
The level of genericity as w e l l a s t h e l e v e l o f abstraction are key elements when choosing a hardware platform and a programming framework in order to solve a concrete problem.
Hardware
SoftwareGeneric
Generic
Specific
Specific
• Arduino/XBee • EPIC/TinyOS
• Campbell RF
Tradeoff : Flexibi l i ty vs . Enforced Reliability